[Mlir-commits] [mlir] [mlir][ROCDL] Add tensor load and store instructions to ROCDL (PR #165016)

[Alan Li]

================
@@ -663,6 +663,50 @@ def ROCDL_GlobalLoadLDSOp :
   }];
 }
 
+//===---------------------------------------------------------------------===//
+// Tensor load/store intrinsics (available in GFX1250)
+//===---------------------------------------------------------------------===//
+
+// Base class for tensor load/store operations with 4 descriptor groups
+class ROCDL_TensorLDSIntrOp<string mnemonic> :
+  ROCDL_IntrOp<mnemonic, [], [], [], 0, 0, 1, 0, [4], ["cachePolicy"]> {
+  dag args = (ins LLVM_Type:$dgroup0, LLVM_Type:$dgroup1, LLVM_Type:$dgroup2,
----------------
lialan wrote:

The LLVM intrinsics lowering to here only support vector types here: https://github.com/llvm/llvm-project/blob/main/llvm/include/llvm/IR/IntrinsicsAMDGPU.td#L4170-L4203

Should use `LLVM_VectorOf` here?

https://github.com/llvm/llvm-project/pull/165016